Remittances, Value Added Tax and Tax Revenue in Developing ...

HAL Id: halshs-00552222https://halshs.archives-ouvertes.fr/halshs-00552222

Preprint submitted on 5 Jan 2011

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Remittances, Value Added Tax and Tax Revenue inDeveloping Countries

Christian Hubert Ebeke

To cite this version:Christian Hubert Ebeke. Remittances, Value Added Tax and Tax Revenue in Developing Countries.2011. �halshs-00552222�

https://halshs.archives-ouvertes.fr/halshs-00552222

https://hal.archives-ouvertes.fr

CERDI, Etudes et Documents, E 2010.30

1

Document de travail de la série

Etudes et Documents

E 2010.30

REMITTANCES, VALUE ADDED TAX AND TAX REVENUE

IN DEVELOPING COUNTRIES

Christian Hubert Ebeke 1

CERDI-Université d’Auvergne Clermont 1

November 2010

1 Clermont Université, Université d’Auvergne, Centre d’Etudes et de Recherches sur le Développement

International (CERDI), 65 Boulevard François Mitterrand, 63000 CLERMONT-FERRAND. France.

E-mail: [email protected].


2

Summary –. This paper examines the impact of international remittances on both the

level and the instability of government tax revenue in receiving countries. It investigates

in particular whether the presence of a value added tax (VAT) system increases the

benefit of the inflows of remittances in terms of high and less volatile tax revenue ratio.

This is supported by the fact that remittances are largely used for consumption

purposes and contribute to smoothing private consumption. Using a large sample of

developing countries observed over the period 1980-2006, and even after factoring in

the endogeneity of remittances and VAT adoption, the results highlight that remittances

significantly increase both the level and the stability of government tax revenue ratio in

receiving countries that have adopted the VAT.

Keywords: Remittances, VAT, Tax revenue, Tax Revenue Instability

JEL codes: H20, F24, E32, O23


3

1. INTRODUCTION

One of the most important issues faced by developing countries and which is

raised even more seriously today is finding ways to improve the internal mobilization of

domestic resources in order to finance public goods. In a context of degradation of

public finances in the developed world, the high dependency of developing countries on

external development assistance should be addressed and reversed by acting on the

ways to efficiently increase the mobilization of domestic resources. In this vein, looking

at the ways to improve the mobilization of domestic resources and to build fiscal space

in these countries is therefore crucial for the sustainability of public finances and for

economic development in general.

If the trend is a stagnation of foreign aid and other forms of development

assistance to developing countries, it is worth noting that, at the same time, developing

countries receive large amounts of external private transfers, namely migrant

remittances. Remittances constitute a bulk of resources received by developing

countries and some scholars argue they represent an external and stable source of

funding for development (Ratha, 2005). To put some numbers in mind, the level of

remittances has attained 338 billions of US dollars in 2008. And despite the recent world

wide crisis, remittances have shown a stronger comparative resilience than the other

types of financial flows received by developing countries. Moreover, for a number of

countries, remittances represent the most important source of external funding, going

beyond the levels of foreign aid or foreign direct investment (Ratha, 2009).

A number of studies has analyzed the macroeconomic impact of remittances in

terms of growth, poverty, competitiveness and macroeconomic instability in receiving

countries. The result showing that remittances significantly reduce poverty rates is

largely accepted among scholars (Adams, 2005; Gupta et al., 2009). However, the role

played by remittances in enhancing economic growth is still an open debate. If the direct

and linear contribution of remittances to economic growth seems difficult to conclude

(Chami et al., 2009b), their positive effect on economic growth conditioned upon some

factors (the quality of governance and the financial development) is now recognized

(Catrinescu et al., 2009; Giuliano and Ruiz-Arranz, 2009; Singh et al., 2009). On average,


4

remittances also contribute towards reducing the volatility of output and consumption

in the receiving countries (Chami et al., 2009a; Combes and Ebeke, forthcoming).

However, these beneficial macroeconomic effects have a price. Remittances can

appreciate the real exchange rate in countries and therefore reduce their external

competitiveness (Amuedo-Dorantes and Pozo, 2004; Acosta et al., 2009).

In this macroeconomic literature on remittances, little is said about their

consequences on public policy. Two recent papers, however, have questioned the

implications of remittance inflows in terms of the sustainability of public finances and

economic policy management. Abdih et al. (2009) investigated the impact of remittances

on the sustainability of government debt using Lebanese fiscal data. Their main result is

that the inclusion of remittances in the traditional analysis of the sustainability of the

debt alters the amount of fiscal adjustment required to place debt on a sustainable path.

They argue that one of the ways remittances can affect fiscal sustainability is the

increase of the tax base. Even if they are not taxed directly, remittance flows may

indirectly increase the revenue that the government receives from consumption-based

and trade-based taxation since they contribute to a higher consumption of domestic and

imported goods. This idea has been empirically confirmed by Abdih et al. (2010) using

panel data for countries in the Middle East, North Africa and Central Asia.

Another paper on the link between remittances and public finances is Chami et al.

(2008). Using a dynamic general equilibrium model calibrated to match the

characteristics of Chilean economy, they showed that the use of a tax on labor income

has the undesirable effect of making the government rely more on inflation to

appropriate resources as the level of remittances increases. This happens because

remittances decrease the labor supply and consequently the labor tax base. In contrast,

when the government uses consumption taxation, an increase in remittances leads to an

increase in tax revenues through private consumption and the government policy is

relatively less distortionary.

The main conclusions of these three recent papers are twofold. Firstly, migrants’

remittances might increase the government tax revenue even if they are not directly

taxed by the government. Secondly, since remittances enter the receiving economy


5

through familial transfers they indirectly affect fiscal policy and the debt sustainability

through the activities of remittance-receiving households, primarily through their

consumption and saving decisions. Abdih et al. (2009) stressed that it is in this respect

that remittances are different from natural resources, which governments may own and

from which they derive revenue, from foreign and domestic public aid, which directly

enter the government budget constraint, and from private capital flows that enter

directly into the production process.

This paper extends the previous analyses to the whole sample of developing

countries. It investigates the contribution of remittances to both the level and the

stability of government tax revenue. Since the effects of remittances on tax revenue are

essentially indirect (through consumption), this paper therefore examines the role of a

value added tax system in the relationship between remittances and total tax revenue

ratio.

Two main hypotheses are tested. (i) Without taxing remittances directly,

government can enjoy more tax revenue ratios through the VAT system since

remittances are largely used for consumption purposes and since the main tax base for

VAT is consumption. (ii) Given the positive contribution of remittances to output and

private consumption smoothing (Chami et al. 2009a; Bugamelli and Paternò, 2009;

Combes and Ebeke, forthcoming), countries that have adopted a VAT will enjoy less

volatile tax revenue ratios thanks to remittance inflows.

This paper raises at least two important debates. Firstly, it explores a new way

through which the adoption of a VAT could enhance the level of tax revenue. A recent

paper has shown the positive contribution of the VAT to the domestic resource

mobilization and its main conclusion is that the VAT is more efficient for tax revenue

mobilization the more opened the economy is (Keen and Lockwood, 2010). By looking at

another aspect of globalization with remittances, our study complements earlier papers.

Secondly, the study examines whether the presence of a VAT is also stabilizing to the

extent that it enhances the stabilizing contribution of remittances to the tax revenue

ratio. This question seems important to investigate since previous studies have

highlighted that the volatility of government tax revenue is quite large for developing


6

countries (Brun et al., 2006) and leads to an instability of government spending (Lim,

1983; Bleaney et al., 1995; Ebeke and Ehrhart, 2010). To our knowledge, this study is

the first one that combines remittances and the VAT-based taxation system to analyze

both the dynamic of the level and the volatility of government revenue in the developing

world.

To test the two formulated hypotheses, a large sample of developing countries

and panel estimators are retained. Whatever the estimator (OLS with fixed effects, Least

Squares Dummy Variables Corrected estimator, Difference-GMM estimator and System-

GMM estimator), the results do not reject the hypothesis that remittances increase both

the level and the stability of government tax revenue in presence of a VAT. These results

are also robust to an alternative definition of workers’ remittances.

The rest of the paper is organized as follows. Section 2 presents the econometric

models of the relationship between remittances, the level and the volatility of the tax

revenue ratio. The section also discusses the data that are used and the identification

strategy. Section 3 presents and discusses the results and the robustness checks that are

conducted. Section 4 concludes on policy implications.

2. EMPIRICAL ANALYSIS

This section presents the specified econometric models, the methodology that is

implemented and the data that are used.

2.1. Tax revenue ratio equation

The econometric model

The econometric model of the determinants of the tax revenue ratio is similar to

previous studies for the choice of control variables (Gupta, 2007; Mahdavi, 2008 and

Keen and Lockwood, 2010). We add two terms to the traditional model: remittances and

remittances interacted with a dummy variable for the VAT. More precisely, the

specification takes the following form:

titititititititi VVRRXty,,,,2,1,,

εηµλθθβα ++++∗++′+= (1)


7

where tity,

is the total tax revenue ratio excluding grants (divided by country GDP), R is

the remittance ratio (in percentage of GDP) and V a dummy variable which takes the

value 1 if there is a VAT in each country i for each year t and 0 otherwise. iµ , t

η and ti ,ε

are the country-specific effect, time-specific effect and the error term, respectively. X is

the set of control variables. We control for the share of agriculture in GDP that is

expected to be negatively correlated with the revenue ratio. It may also serve as a broad

indicator of informality and economic development. Openness (measured as the sum of

the GDP shares of imports and exports) is also a candidate: Rodrik (1998), for example,

finds openness to be positively related to the size of government.2 We also allow

demographic variables–the proportions of the population aged 14 or younger and 65 or

over – to play a potential role. These variables may affect the need for the tax revenue to

support those out of the labor market. Finally, we control for macroeconomic instability

via the inflation rate (measured as the growth rate of the GDP deflator).

In equation (1), the hypothesis tested is that 01

≤θ and 02

>θ so that the impact

of remittances in presence of a VAT (21

θθ + ) is positive while the impact of remittances

on tax revenue ratio in absence of a VAT (1

θ ) is negative or null.3

Equation (1) is firstly estimated by ordinarily least squares with country and time

fixed-effects (OLS-FE). However, this method has important shortcomings in our

context. It does not allow to take into account the dynamic properties of the dependent

variable (the tax rate) nor does it deal with the endogeneity of some regressors.4 The

endogeneity of remittances is of concern here given that altruistic migrants can send

more remittances in order to maintain the purchasing power of their family in their

country of origin, especially when the tax burden is too high and reduces the disposal

2 Given the importance to the VAT of collection at border points, one might also expect this variable to

influence the VAT adoption decision and therefore reduces the selection bias behind the VAT adoption.

3 The impact of remittances in a context of no VAT is supposed to be negative or null since remittances can

reduce labor force participation and therefore reduce the revenue collected from direct taxes (Chami et

al., 2008).

4 Indeed, the tax revenue ratio seems strongly autoregressive year by year given the slow dynamic of

changes in the official tax rates in countries.


8

income. The endogeneity of remittances can also arise in the case of omitted variables.

For example, the emigration of individuals can directly reduce the labor tax base and

therefore total tax revenues, and at the same time, it determines the amount of

remittance inflows.

The endogeneity of the VAT adoption is due to the main motivation behind this

reform. Countries choose to adopt a VAT in order to increase the efficiency of the

revenue mobilization. Therefore, the VAT adoption is highly predetermined.

To deal with these important issues, we proceed in two steps. Firstly, we adopt a

dynamic panel specification of the equation (1). By introducing the lagged value of the

tax revenue ratio, we both control for the inertia of the dependent variable and expect

reducing the self-selection bias in the VAT adoption. The equation takes the following

form:

tititititititititi VVRRXtyty,,,,2,1,1,,

εηµλθθβρα ++++∗++′++= − (2)

Given that the OLS-FE estimator is biased due to the presence of both the lagged

dependent variable and the country fixed-effects, we use the Least Squares Dummy

Variables Corrected estimator (LSDVCE) a method recently proposed by Kiviet (1995),

Judson and Owen (1999), Bun and Kiviet (2003), and extended by Bruno (2005), to

unbalanced panels such as the one used in this study. This method corrects the bias

associated with dynamic panel data model with fixed-effects. The procedure has to be

initialized by a consistent estimator to make the correction feasible, since the bias

approximation depends on the unknown population parameters. In this study, we

initialize the bias correction with the OLS-FE estimator. However, the main drawback is

that the estimated asymptotic standard errors may provide poor approximations in

small samples, possibly generating unreliable t-statistics. The statistical significance of

the LSDVCE coefficients is therefore tested using bootstrapped standard errors (with

100 replications).

The second strategy that we adopt to test the robustnessof our results is to

control for both the dynamic properties of the tax revenue ratio and the endogeneity of

the regressors. Hence an econometric strategy based on instrumental variables must be


9

implemented. Equation (2) in level and equation (2) in first differences are combined in

a system and estimated with an extended GMM estimator system which allows the use

of lagged differences and lagged levels of the explanatory variables (remittance terms

and VAT dummy) as instruments (Blundell and Bond, 1998).5

In order to improve the quality of the instrumentation of the VAT dummy, we

augment standard GMM estimation techniques by adding an external instrument for the

VAT dummy: for each country, the external instrument is the lagged share of

geographical neighbors that have already adopted the VAT for each year t.6 The lagged

value of this variable is used instead of its current value because imitation and

neighborhood effects take time. This variable is also interacted with the lagged values of

remittance ratio to instrument the interactive term of remittances crossed with the VAT

dummy.

For robustness checks, we also conduct estimations using the difference-GMM

estimator (Arellano and Bond, 1991). In this context, equation (2) is firstly differentiated

to eliminate the fixed-effects and the first-differentiated variables are instrumented by

the lagged value of these variables in level. However, our preferred results are derived

from the System-GMM estimator which is asymptotically more efficient in small

samples.

Data

Tax revenue data are drawn from the IMF Government Financial Statistics

database and from several IMF article IV reports. Data exclude grants and are expressed

at the General Government level.

We follow the World Bank in defining remittances as the sum of workers’

remittances and employees’ compensation(s). Workers’ remittances properly refer to

5 The paper uses the System-GMM estimator developed by Blundell and Bond (1998) for dynamic panel

data with the Windmeijer (2005) correction for finite sample bias.

6 Keen and Lockwood (2010) used a similar variable as an important determinant of the VAT adoption.

They also showed that the lagged value of tax revenue ratio significantly and negatively determines the

adoption of the VAT. Thus, by controlling in equation (2) for the lagged value of the tax revenue ratio, we

reduce the endogeneity of the VAT adoption.


10

current transfers by migrants who are employed and reside in the countries where they

migrated (destination country); employees’ compensation should comprise wages,

salaries and other benefits earned by individuals in countries different from their

resident country (country of origin) and for work performed for and paid by residents of

the destination countries. We use the sum of these two items because for many

developing countries the statistical distinction between the two could be highly

problematic (Bugamelli and Paternò, 2009). However, in order to check the robustness

of the results in sub-section 3.3, we test an alternative measure of remittances by using

the narrower definition (only workers’ remittances).

VAT dummy is constructed according to the information on the dates of the

adoption of the VAT that are provided by Ebrill et al. (2001) and updated. All the data for

the remaining control variables are drawn from World Development Indicators.

2.2. Tax revenue instability equation

Econometric model

The econometric model of the determinants of the instability of tax revenue ratio

is similar to previous studies for the choice of control variables (Lim, 1983; Bleaney et

al., 1995 and Ebeke and Ehrhart, 2010). We add two terms to the traditional models:

remittances and remittances interacted with a dummy variable for the presence of a

VAT. More precisely, the specification takes the following form:

tiiiiiiiii VVRRX,,,,2,1,1,,

εηµθγγβφσασ ττττττττ ++++∗++′++= − (3)

where τσ,i is the standard deviation of the growth rate of the total tax revenue ratio

(excluding grants) over the sub-period τ . R is the remittance ratio (in percentage of

GDP) and V a dummy variable which takes the value 1 if there is a VAT in each country i

for at least 3 years in a sub-period τ . iµ , τη and τε

,i are the country-specific effect,

time-specific effect and the error term, respectively. X is the set of control variables,

which includes the standard determinants of macroeconomic volatility such as trade

openness and the standard deviation of the inflation growth rate (inflation volatility).

The hypothesis tested is 02

<γ and ( ) 021

<+ γγ so that the impact of remittances on the


11

instability of tax revenue is more negative in the presence of a VAT than without a VAT

(1

γ ).

We do not control additively for the volatility of the GDP per capita nor for the

volatility of household consumption because they represent the main channels for

remittances to stabilize the tax revenue ratio. However, to test the fact that the

stabilization of private consumption represents the channel for remittances to stabilize

the tax revenue ratio in a country that has adopted a VAT, we add to equation (3) the

standard deviation of household consumption per capita growth rate. If this channel

works, we would get a reduction in the magnitude (in absolute value) and in the

significance of the coefficient of the interaction of remittances crossed with the VAT

dummy.

The time period of this panel consists of non-overlapping sub-periods of 5 years

defined as follows: 1980/1984, 1985/1989… 2000/2005. The variables defined as

standard deviation are computed over each of these sub-periods while the other

remaining variables are defined in terms of sub-period averages.

The dynamic specification is retained to catch the inertia of the instability of

government tax revenue in developing countries. OLS-FE applied to equation (3) leads

to biased estimates and we therefore retain the system-GMM estimator.

Endogeneity issues are still of concern here. Regarding remittances, the

endogeneity can be explained by the fact that they tend to increase in case of negative

shocks that generate macroeconomic instability and therefore government revenue

instability. Regarding the VAT, if governments choose this taxation system in order to

improve the stability of their tax revenues, there is therefore a self-selection bias. By

controlling for the lagged value of tax revenue instability, we partially reduce this bias.

However, we will follow the same identification strategy built in the system-GMM

framework by instrumenting remittances and the VAT dummy by their lagged values

and the share of geographical neighbors with a VAT, respectively. The interactive term

of remittances crossed with the VAT dummy is instrumented by the product of lagged

values of remittances and the share of geographical neighbors that have adopted the

VAT.


12

Except for the tax revenue data, all the other variables used in equation (3) are

drawn from the World Bank Tables. Descriptive statistics of all the variables and the list

of countries included in the sample are reported in the appendix.

3. ESTIMATION RESULTS

The first set of results concerns the impact of remittances on the level of tax

revenue ratio. The second set of results describes the contribution of remittances to the

stabilization of government tax revenue ratio.

3.1. Remittances, VAT and the tax revenue ratio

Table 1 presents the estimations of the impact of remittances on the tax revenue

ratio. For each estimator used, the Table informs about the linear impact of remittances

and, in every second column, about the impact conditional on the presence of the VAT.

Whatever the estimator used, the results indicate that, on average, remittances do

not have a significant impact on the tax revenue ratio. This can be explained by two

effects: on the one hand, remittances may increase tax revenue by expanding the private

demand (consumption of tradable or non-tradable goods, domestic investment). On the

other hand, remittances could reduce the labor supply and increase leisure time at home

(Chami et al., 2005). It results a contraction of the domestic production and therefore in

a sharp decrease in tax revenue ratios for countries that exclusively depend on

corporate and individual taxes (Chami et al., 2008) but not on VAT.

Table 1. about here

The story changes when remittances are interacted with the VAT dummy.

Whatever the estimator used, the interactive term of remittances is statistically

significant and exhibits a positive sign. It is important to keep in mind that the

coefficient of the additive term of remittances identifies the impact of remittances in

countries without a VAT. The sum of the two coefficients associated with remittances

gives the impact for countries that have a VAT system. It emerges that the contribution

of remittances in countries without a VAT is never statistically significant. This is

consistent with the previous papers already quoted. In contrast, remittances seem to


13

significantly increase the tax revenue ratio when countries have implemented a VAT

system.7

When we turn to our preferred estimations by the system-GMM method

(columns [1.7] and [1.8]), the results indicate that the impact of remittances in country

with a VAT stands at 0.08. Perhaps a better sense of the quantitative significance of this

coefficient can be obtained from the following calculation. The median share of the

remittance ratio in our sample is around 1.8 %GDP. A shift from the median toward the

75th percentile of the distribution of the remittance ratio (an increase from 1.8 to 5.3

%GDP) which corresponds to a variation of 3.5 percentage points of GDP would lead to

an increase in the total tax revenue ratio by 0.3 %GDP in a country that has a VAT

system.

3.2. Remittances, VAT and the instability of the tax revenue ratio

Table 2 presents the estimation of the impact of remittances on the instability of

the tax revenue ratio. The estimator used is the system-GMM since remittances are

taken as endogenous and the instability of tax revenue appears strongly autoregressive

in the sample. Column [2.1] reports the linear (unconditional) impact of remittances on

the instability of the tax revenue ratio. The results indicate a non significant stabilizing

effect of remittance inflows.

In column [2.2], the VAT dummy is introduced additively and in interaction with

the remittance ratio. Three important results emerge. Firstly, the presence of a VAT

appears stabilizing. Countries that have adopted a VAT (and whatever their

characteristics, such as openness, the level of development, governance quality or

remittance levels), enjoy more stable tax revenue than the others. This result

complements early studies on the fiscal impact of the VAT adoption. For instance, Keen

and Lockwood (2010) showed that the VAT adoption is a factor enhancing the efficiency

of the tax revenue collection.8 Our results indicate that the VAT is not only a determinant

7 Table 1 reports the sum of the two coefficients associated with remittances and this sum is always

positive and statistically significant in the case of OLS-FE, LSDVC and System-GMM results.

8 More precisely, they showed that the positive effect of the VAT on the tax revenue is conditional on the

degree of trade openness and the level of economic development.


14

of the level of the tax revenue ratio, but also a tax revenue stabilizer. Since the VAT is

based on a macroeconomic aggregate (consumption) that is relatively more stable,

countries that rely on this tax instrument would have less volatile tax revenue. This

result posits the VAT as an effective determinant of overall fiscal performance

(measured as the level of mobilization of revenue and the stability of these revenues) in

developing countries.

Table 2. about here

Secondly, the additive term of remittances which measures the impact of these

flows on the instability of tax revenue in countries without a VAT system, appears not to

be statistically significant although it is negative. This is in line with our expectations.

Indeed, if remittances stabilize private consumption over time, countries that do not tax

consumption via a VAT system would not take advantage of the stabilizing effects of

remittance inflows on private consumption.

Thirdly, the results of column [2.2] indicate a stabilizing effect of remittances on

tax revenue in countries with a VAT. The impact is negative (the sum of the two

remittances coefficients) and stands around -0.28.9 In column [2.3], we test whether this

stabilizing contribution of remittances passes through the stabilization of private

consumption. Hence, the model includes the instability of the household consumption

per capita growth. As expected, its effect on the tax revenue instability is strongly

positive and significant. The inclusion of this variable, however, deteriorates the

significance of the remittance coefficients. This confirms our guess that one of the main

channels through which remittances could stabilize government tax revenue in

countries with a VAT system is through their consumption smoothing properties.

3.3. Robustness checks

One of the main difficulties that arise in the macroeconometric works on

remittances is the measurement error in the remittance variable. Remittances data are

essentially underreported for many countries due to the existence of informal channels.

9 In Table 2, the results of the Wald-test of the joint significance of the two remittance coefficients indicate

a significant effect at almost 5%.


15

Another concern is the fact that many empirical papers that used World Bank data

define remittances as the sum of workers’ remittances and employees’ compensations.

The main justification is that, for many countries, the distinction between workers’

remittances and employees’ compensations is difficult to make. Nevertheless, we test

the robustness of all the previous results, with an alternative definition of remittances

by only using the narrower definition. We define remittances in this section as current

transfers by migrants who are employed and resident in the countries where they

migrated (destination country). Data are drawn from the IMF Balance of Payments

Yearbook (various editions). As previously, remittance values are normalized by country

GDP.

Table 3 reports the results of the impact of remittances on the tax revenue ratio.

The results appear broadly the same than those of Table 1. When remittances are

included additively, and whatever the estimator (OLS-FE, GMM), they do not exhibit a

significant coefficient. In contrast, the conditional effect of remittances on the tax

revenue ratio is statistically significant. Indeed, the interaction term of remittances is

positive and significant, and this result confirms the proposition that remittances tend to

significantly increase the tax revenue ratios in countries that have already adopted a

VAT system. The contribution of remittances in countries with a VAT stands around 0.05

(column [3.6] in the case of system-GMM estimations).

Table 3. about here

Table 4 reports the results of the impact of remittances on the instability of the

tax revenue ratio. The results remain qualitatively the same as those of Table 2. The

inclusion of remittances additively doesn’t lead to any stabilizing effect (column [4.1]).

However, when the model enables to include the interactive term of remittances crossed

with the VAT dummy, the stabilizing contribution of remittances is observed. The value

of the coefficient of remittances for countries that have adopted the VAT (the sum of the

two remittance coefficients) stands at -0.38 (column [4.2]), a value close to that of Table

2 (-0.28). In column [4.3], the inclusion of the instability of consumption per capita leads

to a deterioration of the significance of the coefficients of interest. The consumption

smoothing channel is then empirically confirmed by the data.


16

Table 4. about here

4. CONCLUDING REMARKS

This paper showed robustly that remittances can increase both the level and the

stability of government tax revenue in receiving countries. However, these positive

effects on fiscal performance are only conditional on the presence of the value added tax

system (VAT) in the remittance dependent countries. Even after factoring in the

endogeneity of remittances and the adoption of the value added tax, or even after using

an alternative measure of remittances, the results still remained robust.

The results interestingly add to the positive link between remittances and debt

sustainability analysis in receiving countries (Abdih et al., 2009). They are important for

at least two reasons. Firstly, they showed that public authorities can take advantage of

remittance inflows without taxing them directly, and without creating distortions and

reducing incentives to remit. The presence of a VAT is therefore useful to capture the

positive effect of remittances on consumption smoothing. Secondly, we got clear policy

implications: the adoption of a VAT and measures to increase its effectiveness in

remittance dependent countries can help build fiscal space given the significant

coefficient of transformation of remittances into tax revenue in countries that have

adopted a VAT system.


17

REFERENCES

Abdih, Y., Barajas, A., Chami, R. and Ebeke, C. (2010). Determinants and Fiscal Impacts of

Remittances in the Middle East and Central Asia countries. IMF Working Papers

(forthcoming).

Abdih, Y., Chami, R., Gapen, M. T. and Mati, A. (2009) Fiscal sustainability in remittance-

dependent economies, IMF Working Papers 09/190, International Monetary

Fund.

Acosta, P., Baerg, N. and Mandelman, F. (2009) Financial development, remittances, and

real exchange rate appreciation, Economic Review, pp. 1-12.

Adams, R. and Page, J. (2005) Do international migration and remittances reduce

poverty in developing countries?, World Development, 33, 1645-1669.

Amuedo-Dorantes, C. and Pozo, S. (2004) Workers' remittances and the real exchange

rate: a paradox of gifts, World Development, 32, 1407-1417.

Arellano, M. and Bond, S. (1991) Some tests of specification for panel data: Monte Carlo

evidence and an application to employment equations, The Review of Economic

Studies, pp. 277-297.

Bleaney, M., Gemmell, N. and Greenaway, D. (1995) Tax revenue instability, with

particular reference to sub-Saharan Africa, Journal of Development Studies, 31,

883-902.

Blundell, R. and Bond, S. (1998) Initial conditions and moment restrictions in dynamic

panel data models, Journal of econometrics, 87, 115-143.

Bruno, G. (2005) Approximating the bias of the LSDV estimator for dynamic unbalanced

panel data models, Economics Letters, 87, 361-366.

Bun, M. J. G. and Kiviet, J. F. (2003) On the diminishing returns of higher-order terms in

asymptotic expansions of bias, Economics Letters, 79, 145-152.

Catrinescu, N., Leon-Ledesma, M., Piracha, M. and Quillin, B. (2009) Remittances,

Institutions, and Economic Growth, World Development, 37, 81-92.

Chami, R., Barajas, A., Montiel, P., Fullenkamp, C. and Gapen, M. T. (2009a) Do workers'

remittances promote economic growth?, IMF Working Papers 09/153,

International Monetary Fund.


18

Chami, R., Fullenkamp, C. and Jahjah, S. (2005) Are immigrant remittance flows a source

of capital for development?, IMF Staff Papers, 52, 55-82.

Chami, R., Gapen, M., Barajas, A., Montiel, P., Cosimano, T. and Fullenkamp, C. (2008)

Macroeconomic Consequences of Remittances, IMF Occasional Paper 259, pp. 1-

92.

Chami, R., Hakura, D. and Montiel, P. (2009b) Remittances: An automatic output

stabilizer?, IMF Working Paper 09/91, International Monetary Fund.

Combes, J.-L. and Ebeke, C. (forthcoming) Remittances and household consumption

instability in developing countries, World Development.

Ebeke, C. and Ehrhart, H. (2010) Tax revenue instability in Sub-saharan Africa:

Consequences and remedies, CERDI Working Papers 2010:25, CERDI.

Ebrill, L., Keen, M., Bodin, J. and Summers, V. (2001) The modern VAT, International

Monetary Fund.

Giuliano, P. and Ruiz-Arranz, M. (2009) Remittances, financial development, and growth,

Journal of Development Economics, 90, 144-152.

Gupta, A. S. (2007) Determinants of tax revenue efforts in developing countries, IMF

Working Papers 07/184, International Monetary Fund.

Gupta, S., Pattillo, C. A. and Wagh, S. (2009) Effect of remittances on poverty and

financial development in Sub-saharan Africa, World Development, 37, 104-115.

Judson, R. and Owen, A. (1999) Estimating dynamic panel data models: a guide for

macroeconomists, Economics Letters, 65, 9-15.

Keen, M. and Lockwood, B. (2010) The value added tax: Its causes and consequences,

Journal of Development Economics, 92, 138-151.

Kiviet, J. (1995) On bias, inconsistency, and efficiency of various estimators in dynamic

panel data models, Journal of econometrics, 68, 53-78.

Lim, D. (1983) Instability of government revenue and expenditure in less developed

countries, World Development, 11, 447-450.

Mahdavi, S. (2008) The level and composition of tax revenue in developing countries:

Evidence from unbalanced panel data, International Review of Economics &

Finance, 17, 607-617.


19

Ratha, D. (2005) Workers' remittances: an important and stable source of external

development finance, Remittances: Development Impact and Future Prospects, p.

19.

Ratha, D. (2009). Remittances: A lifeline to poor countries. Finance and Development,

December 2009. International Monetary Fund.

Rodrik, D. (1998) Why do more open economies have bigger governments?, Journal of

Political Economy, 106, 997-1032.

Singh, R., Haacker, M. and woo Lee, K. (2009) Determinants and macroeconomic impact

of remittances in Sub-saharan Africa, IMF Working Papers 09/216, International

Monetary Fund.

Windmeijer, F. (2005) A finite sample correction for the variance of linear efficient two

step GMM estimators, Journal of econometrics, 126, 25-51.


20

Table 1: Remittances, VAT and the tax revenue ratio.

Period:

Unit of observation:

1980-2006

Years

OLS-FE LSDVCE Difference-GMM System-GMM

[1.1] [1.2] [1.3] [1.4] [1.5] [1.6] [1.7] [1.8]

Remittances (% GDP) 0.006 -0.021 0.005 -0.003 0.128 0.057 -0.025 -0.033

0.11 0.53 0.56 0.67 1.05 0.76 0.23 0.56

Remittances * VAT 0.116** 0.042*** 0.165* 0.116**

2.21 6.52 1.69 2.44

VAT dummy -0.859 -0.032 -3.050** -1.631

1.12 0.54 2.13 1.47

Tax revenue % GDP (t-1) 0.759*** 0.730*** 0.656*** 0.593*** 0.636*** 0.638***

46.66 69.22 7.95 6.80 7.78 9.86

Agriculture value added (% GDP) -0.208*** -0.204*** -0.054*** -0.053*** -0.238* -0.471** -0.253** -0.341***

3.53 3.63 4.33 8.26 1.95 2.57 2.27 2.80

Trade openness 0.069*** 0.071*** 0.030*** 0.032*** -0.053 -0.052 0.018 0.025

3.05 3.16 9.12 18.88 1.29 1.25 0.83 1.35

Population aged 14- (%) 0.076 0.073 0.074*** 0.092*** 0.035 0.037 0.290* 0.364**

0.50 0.47 3.76 8.27 0.26 0.21 1.77 2.24

Population aged 65+ (%) -0.417 -0.506 0.230* 0.248*** 0.749 -0.361 0.636** 0.709***

0.58 0.78 1.77 3.80 1.36 0.41 2.24 2.58

log (1+Inflation) -1.109 -1.107 -0.084 -0.114 0.207 -0.117 -1.079* -1.103*

1.22 1.22 0.58 1.56 0.47 0.20 1.73 1.85

Intercept 19.545** 20.135** -1.760 -2.586

2.37 2.56 0.25 0.45

No observations 1702 1702 1640 1640 1526 1492 1640 1600

No countries 98 98 98 98 95 95 98 95

Ho: θ1+θ2=0, p-value 0.093 0.000 0.232 0.016

θ1+θ2 0.09 0.04 0.222 0.083

First order serial correlation p-value 0.000 0.000 0.000 0.000

Second order serial correlation p-value 0.301 0.271 0.204 0.177

Hansen OID test p-value 0.114 0.214 0.095 0.251

No instruments 17 22 15 20

Note: Absolute t-statistics are presented below the corresponding coefficients. Symbols ***, ** and * means significant at 1%, 5% and at 10%. Dependent variable: Total

tax revenue (excluding grants) in percentage of GDP. LSDVCE means the Least Squares Dummy Variable Corrected Estimator initialized by the OLS-FE estimator. In

the system-GMM and difference-GMM framework, the additive term of remittances is taken as endogenous and instrumented by its lagged values. The VAT dummy and

the interactive term remittances*VAT are instrumented by the proportion of geographical neighbors with a VAT and by this variable crossed with lagged values of


21

remittances, respectively. Two-steps GMM estimator with the Windmeijer (2005) correction for finite sample bias is implemented.


22

Table 2: Remittances, VAT and the instability of the tax revenue ratio.

Period 1980-2005

Unit of observation: Non-overlapping 5-year averages

[2.1] [2.2] [2.3]

Remittances (%GDP) 0.425 -0.038 -0.004 1.02 0.36 0.04 Remittances * VAT -0.245* -0.185 1.72 1.53 VAT dummy -4.525* -2.377 1.79 1.07 Instability of tax revenue (t-1) 0.205 0.297** 0.202** 1.52 2.51 2.29 Trade openness -0.027 -0.007 -0.025* 1.36 0.46 1.74 Instability of inflation rate 0.006*** 0.006*** 0.006*** 7.42 15.58 14.49 Instability of private consumption 0.194** 2.16 Intercept 9.366*** 10.075*** 9.738*** 4.76 5.37 5.45 No observations 269 241 235 No countries 96 89 87 Ho: γ1+γ2=0, p-value 0.055 0.161 γ1+γ2=0 -0.283 First order serial correlation p-value 0.123 0.093 0.108 Second order serial correlation p-value 0.275 0.301 0.374 Hansen OID test p-value 0.215 0.460 0.387 No instruments 14 20 21 Note: Absolute t-statistics are presented below the corresponding coefficients. Symbols ***,

** and * means significant at 1%, 5% and at 10%. Dependent variable: Instability of total

tax revenue ratio growth rate (excluding grants). The VAT dummy takes the value 1 if

there is a VAT in at least 3 years in a considered sub-period and 0 elsewhere. In the system-

GMM and difference-GMM framework, the additive term of remittances is taken as

endogenous and instrumented by its lagged values. The VAT dummy and the interactive

term remittances*VAT are instrumented by the proportion of geographical neighbors with

a VAT and by this variable crossed with lagged values of remittances, respectively. Two-

steps GMM estimator with the Windmeijer (2005) correction for finite sample bias is

implemented. Time dummies are included in all specifications.


23

Table 3: Remittances (excluding compensations of employees), VAT and the tax revenue ratio.

Period 1980-2006

Unit of observation: Years

OLS-FE Difference-GMM System-GMM

[3.1] [3.2] [3.3] [3.4] [3.5] [3.6]

Remittances (% GDP) 0.115 0.034 -0.473 -0.059 -0.131 -0.070

1.56 0.39 1.44 1.15 1.43 1.27

Remittances * VAT 0.160* 0.145** 0.122**

1.84 2.07 2.05

VAT dummy -1.512** -1.600** -1.795**

2.10 2.32 2.26

Tax revenue % GDP (t-1) 0.501*** 0.595*** 0.612*** 0.598***

3.88 6.99 10.62 11.81

Agriculture value added (% GDP) -0.212*** -0.217*** -0.401** -0.231** -0.186 -0.302***

4.70 5.10 2.17 2.51 1.49 2.76

Trade openness 0.086*** 0.091*** 0.037** 0.044*** 0.045*** 0.034***

4.10 4.40 2.16 3.17 3.83 2.65

Population aged 14- (%) 0.133 0.102 0.251* 0.154 0.255 0.322***

0.83 0.65 1.75 1.54 1.58 2.89

Population aged 65+ (%) -0.861 -0.908 0.206 -0.124 0.608** 0.645***

1.01 1.18 0.36 0.40 2.42 2.98

log (1+Inflation) -1.131 -1.160 -0.465 -0.367 0.256 0.224

1.36 1.41 0.72 0.63 0.58 0.53

Intercept 17.739* 19.707** -3.071 -1.390

1.89 2.24 0.73 0.39

No observations 1746 1746 1558 1530 1674 1645

No countries 96 96 96 96 96 96

Ho: θ1+θ2=0, p-value 0.025 0.100 0.114

θ1+θ2 0.194 0.086 0.052

First order serial correlation p-value 0.003 0.000 0.000 0.000

Second order serial correlation p-value 0.324 0.218 0.280 0.264

Hansen OID test p-value 0.688 0.350 0.147 0.368

No instruments 12 20 12 16 Note: Absolute t-statistics are presented below the corresponding coefficients. Symbols ***, ** and * means significant at 1%, 5% and at 10%. Dependent

variable: Total tax revenue (excluding grants) in percentage of GDP. In the system-GMM and difference-GMM framework, the additive term of

remittances is taken as endogenous and instrumented by its lagged values. The VAT dummy and the interactive term remittances*VAT are instrumented by


24

the proportion of geographical neighbors with a VAT and by this variable crossed with lagged values of remittances, respectively. Two-steps GMM estimator

with the Windmeijer (2005) correction for finite sample bias is implemented.


25

Table 4: Remittances (excluding compensations of employees), VAT and the instability of

the tax revenue ratio.

Period 1980-2005

Unit of observation: Non-overlapping 5-year averages

[4.1] [4.2] [4.3]

Remittances (%GDP) -0.125 0.391 0.283

0.64 1.16 0.63

Remittances * VAT -0.775** -0.524

2.07 1.16

VAT dummy -1.252 -2.677

0.43 1.00

Instability of tax revenue (t-1) 0.240** 0.168** 0.168**

2.12 1.96 2.51

Trade openness 0.061 0.089 0.023

0.72 0.96 0.23

Instability of inflation rate 0.006*** 0.006*** 0.006***

6.72 10.26 7.56

Instability of private consumption 0.186*

1.69

Intercept 6.014 4.607 7.268

1.29 0.92 1.60

No observations 269 252 243

No countries 93 88 85

Ho: γ1+γ2=0, p-value 0.024 0.267

γ1+γ2=0 -0.384

First order serial correlation p-value 0.114 0.146 0.133

Second order serial correlation p-value 0.248 0.305 0.400

Hansen OID test p-value 0.773 0.761 0.796

No instruments 14 19 20

Note: Absolute t-statistics are presented below the corresponding coefficients. Symbols ***,

** and * means significant at 1%, 5% and at 10%. Dependent variable: Instability of total

tax revenue ratio growth rate (excluding grants). The VAT dummy takes the value 1 if

there is a VAT in at least 3 years in a considered sub-period and 0 elsewhere. In the system-

GMM and difference-GMM framework, the additive term of remittances is taken as

endogenous and instrumented by its lagged values. The VAT dummy and the interactive

term remittances*VAT are instrumented by the proportion of geographical neighbors with

a VAT and by this variable crossed with lagged values of remittances, respectively. Two-

steps GMM estimator with the Windmeijer (2005) correction for finite sample bias is

implemented. Time dummies are included in all specifications.


26

DATA APPENDIX

Summary statistics :

Annual observations

Variables Obs. Mean Std dev. Minimum Maximum

Tax revenue % GDP 2341 20.38 9.03 1.30 70.90

Remittances and Compensations % GDP 2463 4.74 8.35 0 90.42

Remittances % GDP 2561 2.81 5.23 0 41.51

VAT dummy 3888 0.38 0.48 0 1

Percentage of other VAT adopters in the region 3888 37.68 28.59 0 100

Agriculture value added share (%) 3104 23.18 14.51 1.61 93.98

Trade openness (%) 3181 75.48 39.62 1.53 280.36

Population aged 65+ (%) 3648 4.89 2.89 1.89 17.32

Population aged 14- (%) 3648 38.14 8.44 13.57 51.92

log (100+inflation rate) 3295 0.20 0.47 -0.34 5.59

5-year averaged observations

Tax revenue instability 427 13.38 13.95 0.52 195.78

Remittances and Compensations % GDP 483 4.52 8.12 0 78.68

Remittances % GDP 514 2.67 4.86 0 33.15

VAT dummy 720 0.35 0.48 0 1.00

Percentage of other VAT adopters in the region 720 35.16 27.14 0 89.58

Trade openness (%) 606 74.70 38.36 2.35 226.87

Inflation instability 619 104.00 655.15 0.50 11158.63

Instability of real private consumption per

capita growth rate 550 8.44 27.05 0.37 617.30

Countries in the sample (98): Albania, Algeria, Angola, Argentina, Armenia, Azerbaijan, Bangladesh,

Belarus, Belize, Benin, Bolivia, Botswana, Brazil, Bulgaria, Burkina Faso, Burundi, Cambodia, Cameroon,

Cape Verde, Central African Rep., Chad, Chile, China, Colombia, Comoros, Congo Rep., Costa Rica, Cote

d’Ivoire, Djibouti, Dominican Rep., Ecuador, Egypt, El Salvador, Eritrea, Ethiopia, Fiji, Gabon, Gambia,

Georgia, Ghana, Guatemala, Guinea, Guinea-Bissau, Haiti, Honduras, India, Indonesia, Iran, Jamaica, Jordan,

Kazakhstan, Kenya, Kyrgyz Rep., Lao PDR, Lebanon, Lesotho, Madagascar, Malawi, Mali, Mauritania,

Mauritius, Mexico, Moldova, Mongolia, Morocco, Mozambique, Namibia, Nepal, Nicaragua, Niger, Nigeria,

Pakistan, Papua New Guinea, Paraguay, Peru, Philippines, Rwanda, Senegal, Sierra Leone, South Africa, Sri

Lanka, Sudan, Suriname, Swaziland, Syrian Arab Rep., Tajikistan, Tanzania, Thailand, Togo, Tonga, Tunisia,

Uganda, Ukraine, Vanuatu, Vietnam, Yemen Rep., Zambia, Zimbabwe.

Remittances, Value Added Tax and Tax Revenue in Developing ...

Documents

Transcript of Remittances, Value Added Tax and Tax Revenue in Developing ...